Acoustic emission pattern recognition in CFRP retrofitted RC beams for failure mode identification

Abstract

The application of fiber reinforced polymer (FRP) composites to repair reinforcement concrete (RC) structures has emerged as a new and viable choice. However, the understanding of the durability and long-term performance of this combined system still remains elusive. Adopting non-destructive techniques such as acoustic emission (AE) will raise confidence in exploiting the full potential of this material. The objective of the current study is to identify failure mechanisms in CFRP-retrofitted RC beams by applying advanced pattern recognition techniques on the collected AE data. Six RC beams with artificially induced damage repaired with CFRP sheets are tested with flexural loads and monitored with AE sensors. Since damage mechanisms in the retrofitted RC beams are unknown a priori, a pattern recognition methodology is developed. After preprocessing the AE data using the principal component analysis (PCA), the unsupervised k-means clustering method is applied to automatically cluster and separate the AE patterns. The neural networks based on multi-layer perceptron (MLP) or support vector machine (SVM) algorithm are then developed to better understand the trends in the AE data and their association with the observed damage mechanism. Finally, the trained models are used to successfully identify damage modes in other similar samples.